Back
34/65
Lesson 34
6 min
FSFS: Hybrid Local Search
Two-tier lexical + semantic search with progressive delivery
Ready to level up?
Mark complete to track your learning progress.
Two-tier lexical + semantic search with progressive delivery
Mark complete to track your learning progress.
Search your local files at blazing speed with FrankenSearch's hybrid retrieval engine.
FSFS (FrankenSearch) is a two-tier hybrid local search engine combining lexical (BM25) and semantic retrieval with progressive delivery. It indexes your local files and returns results ranked by both keyword relevance and meaning.
FSFS builds an index once, then serves queries in milliseconds. It supports incremental updates so new files are searchable without a full re-index.
Fast keyword matching with TF-IDF scoring
Embedding-based similarity for meaning-aware results
Results stream as they're found, fast first results
Only re-indexes changed files for fast updates
Explore how FSFS combines keyword matching (BM25) with semantic vector search to deliver results that neither approach can achieve alone. Step through different query scenarios and watch the full search pipeline animate.
Compare BM25 vs semantic vs hybrid fusion across 6 scenarios
Exact function name lookup -- BM25 dominates because tokens match perfectly.
Select a scenario and click Run to see the hybrid pipeline in action
# Index a projectfsfs index ~/projects/my-app# Search for error handling patternsfsfs search "error handling retry"# Find conceptually similar codefsfs search --semantic "authentication middleware"